Balancing effluent quality, economical cost and greenhouse gas emissions during the evaluation of plant-wide wastewater treatment control strategies

Xavier Flores-Alsina1, Magnus Arnell1,2 ,Youri Amerlinck3, Lluís Corominas4, Krist V. Gernaey5, Lisha Guo6, Erik Lindblom1,7, Ingmar Nopens3, Jose Porro3,8, Andy Shaw9, Laura Snip5, Peter A. Vanrolleghem6, Ulf Jeppsson1

1Department of Measurement Technology and Industrial Electrical Engineering (MIE), Division of Industrial Electrical Engineering and Automation (IEA), Lund University, Box 118, SE-221 00 Lund, Sweden. 2CIT Urban Water Management, Gjuterigatan 1D, SE-582 73 Linköping, Sweden. 3BIOMATH, Department of Mathematical Modelling, Statistics and Bioinformatics, Ghent University, Coupure Links 653, B-9000 Ghent, Belgium. 4ICRA (Catalan Institute for Water Research), Scientific and Technological Park of the University of Girona, H2O Building, Emili Grahit 101, 17003 Girona, Spain. 5Center for Process Engineering and Technology (PROCESS), Department of Chemical and Biochemical Engineering, Technical University of Denmark, Building 229, DK-2800 Kgs. Lyngby, Denmark. 6modelEAU, Département de génie civil et de génie des eaux, Université Laval, 1065 Avenue de la Médecine, Québec G1V 0A6, QC, Canada. 7Sweco Environment, Gjörwellsgatan 22, SE-100 26 Stockholm, Sweden. 8Malcom Pirnie, The Water Division of ARCADIS, 27-01 Queens Plaza, Ste. 800, Long Island City, NY 11101, USA. 9Black and Veatch, 8400, Ward Parkaway, Kansas City, MO 64114, USA.

Abstract: The objective of this paper is to complement the effluent quality (EQI) and operational cost (OCI) indices used to evaluate (plant-wide) control strategies in wastewater treatment systems with a new dimension dealing with greenhouse gas (GHG) emissions. The GHG evaluation is based on a set of comprehensive dynamic models that estimate the most significant potential on-site (secondary and sludge treatment, sludge disposal) and off-site (net energy use, embedded chemicals) sources of GHG emissions. The case study presented here calculates and discusses the changes in the EQI, OCI and the formation of carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O) as a consequence of varying model assumptions related to four process variables: i) system aeration in the activated sludge (AS) section; 2) percentage of particulates removed in the primary clarifier; 3) the temperature (T) regime in the anaerobic digester; and 4) the control of nitrogen rich returns coming from the sludge treatment. Simulation results show the undesirable effects that energy optimization might have on the GHG production: Even though it decreases off-site CO2 emissions, as a result of: i) reduced aeration energy requirements and ii) increased energy recovery from the sludge treatment, such effects might be counterbalanced by increased N2O emissions in the AS due to the 300-fold stronger greenhouse effect of N2O. The reported results emphasize the importance of a plant-wide approach and the need to consider the interactions between the different treatment units when evaluating the global warming potential (GWP) of a wastewater treatment plant (WWTP). Finally, the paper demonstrates the potential of using the proposed approach as a general model-based tool for determining the most sustainable WWTP operational strategies, which is essential in a water sector where climate change, energy and sustainability, are key challenges to be tackled. Keywords: Activated sludge modelling, Benchmarking, Global warming, Model-based evaluation, Multi-criteria decision making, Process control, Sustainability

1. INTRODUCTION The constantly changing nature of wastewater (quantity/quality), its unknown origin and the great variety of ambient conditions make wastewater treatment plants (WWTPs) truly dynamic systems. Comprehensive studies and full-scale applications (Olsson et al., 2005) have shown the feasibility of using automatic control to optimize the operation under these conditions. WWTP models and simulation studies have been used to evaluate performance and compare control strategies in general (Gernaey et al., 2012) or before full-scale implementation (Ayesa et al., 2006). The complexity of modern WWTPs with different sub-processes, interconnections and recirculations

makes it necessary to consider a plant-wide perspective in order to avoid sub-optimal performance (Olsson and Newell, 1999; Jeppsson et al., 2007; Nopens et al., 2010). The main focus for a WWTP has historically been the effluent water quality under constraints of technical feasibility and cost. This still holds, but the discussion on sustainability in general, and the issue of climate change due to greenhouse gas (GHG) emissions in particular, have widened the scope for the utilities. An increasing interest for GHG emissions calls for new approaches to reach the high and increasing demands on effluent quality and at the same time predict and minimize the GHG emissions. For this reason, the main objective of this paper is to complement the traditional effluent quality index (EQI) and operational cost index (OCI) used to evaluate control strategies in a WWTP with a new dimension dealing with GHG emissions. In the presented case study a modified version of the International Water Association (IWA) Benchmark Simulation Model No 2 (BSM2) is evaluated alongside a number of scenarios. Thus, changes in effluent quality, operational cost and CO2, CH4 and N2O emissions are analyzed in a plant-wide fashion when i) the system aeration in the activated sludge (AS) section; ii) percentage of particulates removed in the primary clarifier (PRIM); 3) the temperature regime in the anaerobic digester (AD); and iii) the control of nitrogen rich returns coming from the sludge line are modified. 2. METHODS 2.1. WASTEWATER TREATMENT PLANT UNDER STUDY The WWTP under study (BSM2G) has the same layout as the IWA BSM2 proposed by Jeppsson et al. (2007) and Nopens et al. (2010). The activated sludge unit is a modified Ludzack-Ettinger configuration consisting of 5 tanks in series. Tanks 1 (ANOX1) and 2 (ANOX2) are anoxic, while tanks 3 (AER1), 4 (AER2) and 5 (AER3) are aerobic. AER3 and ANOX1 are linked by means of an internal recycle. The BSM2G plant further contains a primary (PRIM) and a secondary (SEC) clarifier, a sludge thickener (THK), an anaerobic digester (AD), a storage tank (ST) and a dewatering unit (DW). Further information about the models used can be found in Corominas et al. (2012) and Porro et al. (2011). From the original set of models, the Activated Sludge Model No 1 (ASM1) (Henze et al., 2000) has been expanded with the principles stated in Hiatt and Grady (2008) to take into account N2O formation during denitrification. This model incorporates two nitrifying populations: ammonia oxidizing bacteria (AOB) and nitrite oxidizing bacteria (NOB) using free ammonia and free nitrous acid, respectively, as their substrates. The model also considers sequential reduction of nitrate to nitrogen gas via nitrite, nitric oxide and nitrous oxide using individual reaction specific parameters. Additionally, the ideas summarized in Mampaey et al. (2011) are used to consider NO and N2O formation during the nitrification pathway assuming ammonia as the electron donor. The interfaces presented in Nopens et al. (2009) have been modified to link the modified activated sludge model and the anaerobic digestion model (Batstone et al., 2002), by considering COD, N and charge balances for all oxidized nitrogen compounds. At the activated sludge side five new variables are defined compared to the ASM1 model used in BSM2: NO2

-, NO, N2O, N2 and an additional autotrophic biomass variable (XBA2). Further information about the models used can be found in Corominas et al. (2012) and Porro et al. (2011). 2.2. CONTROL STRATEGY AND SIMULATED SCENARIOS The plant is simulated in closed loop regime, which includes two PI control loops. The first loop controls the dissolved oxygen concentration in AER2 by means of manipulating the air flow rate, here implemented as manipulation of the oxygen transfer coefficient KLa. The second loop controls the nitrate concentration in the 2nd anoxic tank (ANOX2) by manipulating the internal recycle flow

rate (Qintr). Two different waste sludge flow rates (QW = 300 m3.day-1 // QW = 450 m3.day-1) are imposed in SEC depending on the time of the year in order to sustain the nitrifying biomass in the system during the winter period. Noise and delays are applied to sensors and actuators to give the simulations more realism. External recirculation flow rate (Qr) and carbon source addition (Qcarb) remain constant throughout the simulations. Additional details about the operational strategy can be found in Flores-Alsina et al. (2011).Four different control strategies / scenarios are simulated in the presented case study:

Changing the DO set-point value from 1 to 3 g.m-3 (default value 2 g.m-3); Changing the % TSS removal efficiency in PRIM from 33% to 66% (default value 50%); Changing the temperature regime in the AD from mesophilic (35°C) to thermophilic (55°C)

(default value 35°C); Controlling the return flow rate coming from the digester. Storing during day time (when the

plant is highly loaded) and releasing at night (when the plant is low loaded) (default strategy, no control).

2.3. EVALUATION CRITERIA 2.3.1. Effluent quality (EQI) and operational cost (OCI) indices The overall pollution removal efficiency is obtained using the effluent quality index (EQI) from the standard BSM2 (Nopens et al., 2010). EQI is an aggregated index of all the pollution loads TSS, COD, BOD5, total Kjeldahl nitrogen (TKN) and NOX leaving the plant. The economic objectives are evaluated using the operational cost index (OCI) (Nopens et al., 2010). It consists of a weighted sum of all the major operating costs in the plant: aeration energy (AE), pumping energy (PE), mixing energy (ME), sludge production (SP), external carbon addition (EC), methane production (MP) and the net heating energy (HEnet) needed to heat the sludge in the AD. EQI and OCI are based on one-year dynamic influent data Gernaey et al. (2011). 2.3.2. Greenhouse gas (GHG) emissions The comprehensive method proposed by Flores-Alsina et al. (2011) is used to calculate GHG emissions in the WWTP. The emissions considered are: Direct secondary treatment emissions: The emissions from the activated sludge section include the CO2 generated from biomass respiration and BOD oxidation, the N2O generated from nitrogen removal and the CO2 credit from nitrification. Sludge processing: The GHG emissions during sludge treatment are mainly generated in the anaerobic digester. In this case it is assumed that the biogas is fed directly into a gas-fired combustion turbine converting the CH4 into CO2 and generating electricity and heat (in turn used to heat the anaerobic digester). The CO2 generated during anaerobic digestion and the CO2 produced in the combustion process are assumed to be released to the atmosphere. Net power GHG: The difference between energy usage and production. Energy consumption involves aeration, pumping, mixing and heating. Energy production comes from the electricity generated by the turbine. A value of 0.94 kg CO2 per kWh is assumed for any external energy production (based on a coal-burning power plant (Bridle et al., 2008)). Chemicals: The GHG emissions from production of carbon source for denitrification are accounted for (from industrial production of methanol data) (Dong and Steinberg, 1997). Sludge disposal and reuse: The disposal of sludge is accounted for with CO2 emissions from transport and mineralization of organic matter to/at the disposal site. GHG emissions are also evaluated over one year. Finally, in order to deal with the different nature of the generated GHG emissions (CO2, CH4 and N2O) are converted into units of CO2 equivalents (CO2e). The assumed global warming potential (GWP) for N2O and CH4 is 298 kg CO2e per kg N2O and 25 kg CO2e per kg CH4, respectively (IPCC, 2006).

3. RESULTS EQI, OCI and GHG values for the different simulated scenarios are depicted in Figures 1a and b. From the generated results it is possible to see that: 1) dissolved oxygen concentration has a paramount importance on the total GHG emissions; 2) % TSS removal efficiency mainly influences effluent quality and operational cost, but the total GHG emissions remain almost equal; 3) thermophilic conditions in the AD reveal that a higher operating temperature appears to be a more expensive way to operate the plant (with higher GHG emissions) without having substantial benefits in terms of gas production; and, 4) control of return flow rate slightly reduces effluent quality, but it does not have an effect on the GHG emissions unless DO is very low (see dotted lines in Figure 1b). Further details are provided in the following sub-sections. a

b

Figure 1. Effluent quality, operational cost and greenhouse gases emissions for the different control strategies with (a) and without (b) controlling the nitrogen rich returns from the digester

3.1. EFFECT OF DISSOLVED OXYGEN CONCENTRATION Low DO set-points lead to a reduction of the off-site CO2 production due to lower energy consumption (and subsequently operational cost) but the overall GHG emissions are increased compared to the default case. The reason of this GHG increase is due to N2O and its 300-fold stronger greenhouse effect. In this case, the N2O increase is mainly caused by accumulation of NO2

- (see Figure 2b) due to incomplete nitrification (see the increase of the EQI values in Figure 1 and the dynamic performance of NH4

+ in Figure 2a). High DO set-points increase operational costs, but improve effluent quality (see Figure 1). Although off-site emissions of CO2 are higher, the overall GHG emissions are still lower due to reduced N2O emissions. In Figure 2, but also in Figures 3 and 4, the seasonal variation can be seen (starting simulation date: 1st of July, total time: 364 days). a

b

Figure 2. Effect of dissolved oxygen on nitrification: NH4

+ (a) and NO2- (b) in the effluent

3.2. EFFECT OF PRIMARY CLARIFIER EFFICIENCY Although the total GHG emission seems to be very similar (z-axis in Figure 1), types of GHG and origin change substantially in the different simulated scenarios. High PRIM efficiency (TSS removal = 66%) decreases the quantity of TSS entering the activated sludge section leading to better effluent quality (although denitrification is significantly worsened because of lack of carbon source). The lower operational cost is due to: 1) better energy recovery in the sludge line due to an increased biogas production (see Figure 3b); and, 2) lower aeration cost. However, the latter increases the overall N2O emissions due to low C/N ratio as a trade-off (see Figure 3a). In terms of GHG emissions low clarifier efficiency (TSS removal = 33%) causes: 1) an increase of the biogenic CO2 emissions from BOD oxidation and biomass decay in the bioreactor; 2) an increase of off-site CO2 emissions due to higher energy demand during nitrification; 3) a reduction of the energy recovery from settled organics (Figure 3b); and, 4) a decrease of N2O emissions due to a higher C/N ratio (Figure 3a). a

b

Figure 3. Effect of dissolved TSS removal efficiency in PRIM: N2O emissions in AS (a) and CH4 in AD (b)

3.3. EFFECT OF DIGESTER PERFORMANCE Figure 4 shows the results of changing the digestor’s operating temperature from 35 °C (mesophilic conditions) to 55 °C (thermophilic conditions). No substantial benefits can be observed in either gas production or off-site CO2 emissions. However, thermophilic conditions substantially increase the operational cost (see Figure 1) due to higher energy requirements for heating (see Figure 4a) without improving the digester performance (see Figure 4b). Subsequently this also leads to higher CO2 emissions from off site power generation. The effect on effluent quality variables is negligible. a

b

Figure 4. Effect of modified T regime on the AD performance: energy demand (a) and CH4 (b) in AD

3.4. EFFECT OF NITROGEN RICH RETURNS FROM DIGESTION From the results depicted in Figure 1 one can see the effect of controlling the nitrogen rich returns from the AD section. In all cases, there is a slight improvement in the effluent quality (all the evaluated scenarios have lower values on the x-axis). This reduction is attributed to the storage tank’s smoothing effect on the ammonium peaks originating from the sludge treatment line. The ammonium rich return flows are stored during day time when the plant is already highly loaded, and are released during night time when the influent nitrogen load is lower (see N load entering the AS section in Figure 5). The slight increase in the OCI (y-axis) is due to the extra pumping. Finally, when it comes to GHG there is no substantial effect unless the DO concentration is very low. Again, this is caused by a substantial reduction in the N2O emissions, where an oxygen deficit combined with high ammonium loadings can increase its production.

Figure 5. Smoothing effect on the nitrogen load entering the AS section with and without including control (20 day

snapshot of 364 days simulation)

4. DISCUSSION The results reported in this paper reach similar conclusions as the experiments reported in Schulthess and Gujer (1996) and Kampschreur et al. (2009), related to DO, C/N ratios and N2O emissions. There is also a good match with the studies of the effect of soluble/particulate compounds in the AS and the relation with the overall GWP of the plant (Gori et al., 2011). Nevertheless, there are aspects that still need to be addressed. For example, there is evidence that N2O production increases during winter time (Kampschreur et al., 2009). In our case, lower temperatures have the opposite effects. The latter can be explained by the pronounced effect of the temperature on the maximum growth of autotrophic and heterotrophic bacteria in the used model. The authors are aware of the fact that a TSS removal of 66% in PRIM is hard to achieve in many treatment plants without the addition of chemicals (Tchobanoglous et al., 2003). Further research is necessary to consider the role of these chemicals within the operational cost index and the overall GWP in a similar way as is done for carbon source usage, i.e. kg CO2e for each kg of chemical used. The poor digestion improvement when the temperature is changed to thermophilic conditions can be explained by the following points: i) the used ASM/ADM interfaces (Nopens et al., 2009) where the disintegration process (limiting factor in many digestion processes) is instantaneous; ii) the low concentration of soluble organics coming with the influent (Gernaey et al., 2011) and consequently limited amounts of organic material arriving to the AD (although kinetics are faster at thermophilic conditions, there is no more material to be converted) and iii) finally the large digestion volume i.e. sufficiently long hydraulic retention time during mesophilic conditions to convert all the potentially digestible organics into methane.

The case study shows that the presented models are useful to quantify the different GHG emissions when evaluating control strategies/operational procedures taking into account the different sources of CO2, CH4 and N2O. However, from a climate change point of view, not all these sources have the same importance. For example, biogenic sources of CO2 such as the generation from the aerobic/anaerobic treatment processes are part of the natural carbon cycle. On the other hand, there are non-biogenic sources such as the off-site CO2 emissions due to electricity consumption or production of chemicals that should be avoided. A clear example can be found in scenario 2 (% TSS removal efficiency, where the total GHG is almost the same, but the types and origins are quite different). The study intended to show the additional degree of complexity that results from adding another dimension to the traditional evaluation criteria used to compare and evaluate (plant-wide) control strategies in wastewater treatment systems. Simulation of the modified BSM2 showed the existing interactions and trade-offs between effluent quality, economic cost and GHG emissions. For example, it was possible to analyse the side effects of energy optimization, particularly in the aeration system. Low DO levels decrease energy usage, but on the other hand reduce effluent quality. and to a large extent increase GHG emissions substantially due to increased release of N2O. The other example is energy recovery from digestion, where a change in the influent C/N ratio also increases N2O. Even though an overloaded reactor might increase the quantity of biogenic CO2 emissions (BOD oxidation, biomass decay) there is a drastic decrease in N2O emissions. It is important to highlight that some of the models used in the case study are still under development. In this paper, the N2O production is based on AOB denitrification with NO2 as terminal electron acceptor. However, other possible mechanisms, such as the formation of N2O as a by-product of incomplete oxidation of hydroxylamine (NH2OH) to NO2, are omitted. Recent investigations demonstrate that both the denitrification and NH2OH pathways may be involved in N2O production. Unfortunately, a unified model that describes both mechanisms independently does not yet exist (Ni et al., 2012). Finally, the reader should be aware of that the list of emissions applied in this case study is not complete. There are other sources of GHG that potentially contribute to the overall GWP of the plant. Experimental observations have revealed that substantial stripping of methane might take place at the inlet of the WWTP (Guisasola et al., 2009). Also, no fugitive emissions of methane are considered from the anaerobic digester (Czepiel et al., 1993). In the ADM-ASM interfaces (Nopens et al., 2009), the quantity of methane that remains in the liquid phase is stripped, but not quantified in the model. Finally, the N2O and CH4 emissions from sludge disposal and reuse are not considered either (EPA, 2010). 5. CONCLUSIONS The work demonstrates the changes of effluent quality, operational cost and GHG emissions during the simulation-based evaluation of plant-wide wastewater treatment systems. The key findings can be summarized as:

• Low DO levels decrease off-site CO2 emissions and cost (due to aeration) but increase the plant´s GWP (higher N2O emissions) and worsen effluent quality (high NH4

+ values); • High TSS removal efficiency in PRIM improves energy recovery from the AD and the

effluent quality, but increases N2O emissions as well; • Thermophilic conditions in the AD increase cost and GHG emissions without additional

biogas production compared to mesophilic conditions. This is mainly due to some of the model assumptions, influent conditions and digester design

• Control of returns slightly improves effluent quality and GHG emissions when DO levels are low.

ACKNOWLEDGEMENTS The authors fully acknowledge the intellectual support of the IWA task groups on “The use of water quality and process models for minimizing wastewater utility greenhouse gas footprints” and “Benchmarking of control strategies for wastewater treatment plants”. Peter Vanrolleghem holds the Canada Research Chair in Water Quality Modelling. The authors acknowledge the financial support obtained through the Swedish Research Council Formas (211-2010-141), the Swedish Water & Wastewater Association (10-106, 11-106), the Richert Foundation (PIAH/11:58), the TECC project of the Québec Ministry of Economic Development, Innovation and Exports (MDEIE) and FWO-Flanders (G.A051.10). REFERENCES Ayesa E., De la Sota A., Grau P., Sagarna J.M, Salterain A. and Suescun J. (2006). Supervisory control strategies for the new WWTP of Galindo-Bilbao: The long run from the conceptual design to the full-scale experimental validation. Water Sci. Technol., 53(4-5), 193-201. Batstone D.J., Keller J., Angelidaki I., Kayuznyi S.V., Pavlostathis S.G., Rozzi A., Sanders W.T.M., Siegrist H. and Vavilin V.A. (2002). Anaerobic Digestion Model No 1. IWA STR No 13, IWA Publishing, London, UK. Bridle T., Shaw A., Cooper S., Yap K.C., Third K. and Domurad M. (2008). Estimation of greenhouse gas emissions from wastewater treatment plants. In: Proceedings IWA World Water Congress 2008, Vienna, Austria. September 7-12, 2008. Corominas Ll., Flores-Alsina X., Snip L. and Vanrolleghem P.A. (2012). Comparison of different modelling approaches to better understand greenhouse gas emissions from wastewater treatment plants. Biotechnology & Bioengineering (accepted). Czepiel P.M., Crill P.M. and Harris R.C. (1993). Methane emissions from municipal wastewater treatment processes. Env. Sci. Tech., 27, 2472-2477. Dong Y. and Steinberg M. (1997). Hynol: an economical process for methanol production from biomass and natural gas with reduced CO2 emission. Int. J. Hydrogen Energy, 22(10-11), 971-977. EPA (2010). Methane and Nitrous Oxide Emissions from Natural Sources. Technical Report published April 2010. Flores-Alsina X., Corominas L., Snip L. and Vanrolleghem P.A. (2011). Including greenhouse gas emissions during benchmarking of wastewater treatment plant control strategies. Water Res., 45(16), 4700-4710. Gernaey K.V., Flores-Alsina X., Rosen, C., Benedetti, L. and Jeppsson U. (2011). Dynamic influent pollutant disturbance scenario generation using a phenomenological modelling approach. Env. Mod & Soft., 26(11), 1255-1267. Gernaey K.V., Jeppsson U., Vanrolleghem P.A., Copp J.B. and Steyer J.-P. (eds) (2012). Benchmarking of Control Strategies for Wastewater Treatment Plants. IWA Scientific and Technical Report, IWA Publishing, London, UK (to appear). Gori R., Jiang L.M., Sobhani R. and Rosso D. (2011). Effects of soluble and particulate substrate on the carbon and energy footprint of wastewater treatment processes. Water Res., 18(15), 5858-5872. Guisasola A., Sharma K.R., Keller J. and Yuan Y. (2009). Development of a model for assessing methane formation in rising main sewers. Water Res., 43(11), 2874-2884. Henze M., Gujer W., Mino T. and van Loosdrecht M.C.M. (2000). Activated Sludge Models ASM1, ASM2, ASM2d and ASM3. IWA Scientific and Technical Report No 9, IWA Publishing, London, UK. Hiatt W.C. and Grady C.P.L.Jr. (2008). An updated process model for carbon oxidation, nitrification, and denitrification. Water Environ. Res., 80, 2145-2156. IPCC (2006). 2006 IPCC Guidelines for National Greenhouse Gas Inventories. Intergovernmental Panel on Climate Change. Available at: http://www.ipccnggip.iges.or.jp/public/2006gl/index.html. Jeppsson U., Pons M.N., Nopens I., Alex J., Copp J.B., Gernaey K.V., Rosen C., Steyer J.P. and Vanrolleghem P.A. (2007). Benchmark Simulation Model No 2 – General protocol and exploratory case studies. Water Sci. Technol., 56(8), 287-295. Kampschreur M.J., Temmink H., Kleerebezem R., Jetten M.S.M. and van Loosdrecht M.C.M. (2009). Nitrous oxide emission during wastewater treatment. Water Res., 43(17), 4093-4103. Mampaey K.E., Beuckels B., Kampschreur M.J., Kleerebezem R., van Loosdrecht M.C.M. and Volcke E.I.P. (2011). Modelling nitrous and nitric oxide emissions by autotrophic ammonium oxidizing bacteria. In: Proceedings IWA/WEF Nutrient Recovery and Management Conference, Miami, FL, USA, January 9-12 2011. Ni B.J., Yuan Z., Chandran K., Vanrolleghem P.A. and Murthy S. (2012). Evaluating mathematical models for N2O production by ammonia-oxidizing bacteria: towards a unified model. In: Proceedings 3rd IWA/WEF Wastewater Treatment Modelling Seminar (WWTmod2012), Mont-Sainte-Anne, Québec, Canada, February 27-29 2012. Nopens I., Batstone D.J., Copp J.B., Jeppsson U., Volcke E., Alex J. and Vanrolleghem P.A. (2009). An ASM/ADM model interface for dynamic plant-wide simulation. Water Res., 43(7), 1913-1923.

Nopens I., Benedetti L., Jeppsson U., Pons M.-N., Alex J., Copp J.B., Gernaey K.V., Rosen C., Steyer J.-P. and Vanrolleghem P.A. (2010). Benchmark simulation model No 2: Finalisation of plant layout and default control strategy. Water Sci. Technol., 62(9), 1967-1974. Olsson G. and Newell B. (1999). Wastewater Treatment Systems - Modelling, Diagnosis and Control. IWA Publishing, London, UK. Olsson G., Nielsen M.K., Yuan Z., Lynggaard-Jensen A. and Steyer J.P. (2005). Instrumentation, Control and Automation in Wastewater Systems. IWA Publishing, London, UK. Porro J., Guo L., Sharma K., Benedetti L., Van Hulle S., Vanrolleghem P.A., Amerlinck Y., Yuan Z., Shaw A. and Nopens I. (2011). Towards a benchmarking tool for minimizing wastewater utility greenhouse gas footprints. In: Proceedings 8th International IWA Symposium on Systems Analysis and Integrated Assessment in Water Management (Watermatex2011), San Sebastian, Spain, June 20-22 2011, 507-517. von Schulthess R. and Gujer W. (1996). Release of nitrous oxide (N2O) from denitrifying activated sludge: Verification and application of a mathematical model. Water Res., 30(3), 521-530. Tchobanoglous G., Burton F.L. and Stensel H.D. (2003). Wastewater Engineering: Treatment, Disposal and Reuse. McGraw-Hill, New York, USA.